Experimental eddy current measurements of flawed edges compared with results from probabilistic numerical models

Abstract

Eddy current detection of flaws in edges presents challenges in experimental procedures during benchmark studies in the laboratory for model validation as well as practical implementation of a real world detection system. These difficulties result in distortions to the signal that mask the effects from the flawed region itself. Rather than attempting to perfect the experimental setup, we propose to make the numerical models more robust by incorporating randomness in the experimental procedure with uncertainty quantification methods. We present the motivation for the specific method chosen, the probabilistic collocation method (PCM), and the mathematical development behind the method, and then present the results from numerical simulations with a validation measure.